Research archive / public

Evidence review / required

Hyx studies

Systems research from the edge.

Technical studies of how Hyx works. Claims link to source evidence, controlled measurements, or primary literature. Limitations appear alongside the results.

09: published studies; reviewed for public disclosure
100%: evidence mapped; claims linked to an evidence register
0: hidden limitations; scope and uncertainty stated

Published archive

Studies that survived the evidence review.

09 records

Study / 09Published

Network intelligence

11 July 2026

Network intelligence without a network call

A study of normalized IP ranges, flat local tables, verified-bot identity, snapshot replacement, and the uncertainty of reputation data.

Geo and reputation context is useful only if lookup is predictable and provenance remains visible. Hyx compiles external datasets into local range tables, keeps request-time lookup offline, and treats category membership as evidence rather than guilt.

01focused tests passed02verified-bot lookup03lookups / second04remote calls / lookup

Read the study→

Study / 08Published

Representation systems

11 July 2026

Optimize once. Serve the derivative

A study of content negotiation, background image conversion, derivative caching, compression choice, and representation safety at the edge.

Representation optimization is not a single codec benchmark. It is the system that chooses an acceptable form, creates expensive derivatives away from the response path, preserves cache correctness, and refuses unsafe inputs.

01focused tests passed02encoding negotiation03best HTML ratio04output families

Read the study→

Study / 07Published

Edge observability

11 July 2026

One request. One attributable record

A schema and ingestion study of delivery, transport, security, signals, model, and fleet context written together at the edge.

Investigation slows down when identity, enforcement, performance, and model evidence live in different systems. Hyx writes the request context as one wide record, then batches it without allowing sampling to erase mandatory security outcomes.

01joined dimensions02JSONEachRow encode03rows / second04focused tests passed

Read the study→

Study / 06Published

HTTP caching

11 July 2026

Caching is a protocol, not a map

A semantics study of freshness, variants, validators, stale delivery, request coalescing, ranges, and tiered storage beyond eviction policy.

Mobius answers how memory is retained under pressure. This study asks a different question: whether the response cache preserves the HTTP contracts that decide what may be stored, reused, revalidated, ranged, or served stale.

01focused tests passed02no-range fast gate03range parser04storage tiers

Read the study→

Study / 05Published

Transport intelligence

11 July 2026

Identity before HTTP

A study of TLS ClientHello evidence, QUIC visibility, normalized fingerprints, and structural bot classification before application policy runs.

HTTP headers can be forged after the transport is established. Hyx therefore extracts evidence from the handshake and carries it forward as attributable identity context, not absolute identity.

01parse + JA3 + MD502structural classify03reputation refine04focused tests passed

Read the study→

Study / 04Published

Application security

11 July 2026

A WAF is a state machine

A compatibility and execution study of OWASP CRS semantics across request, response, scoring, exclusion, and live configuration boundaries.

Hyx treats Core Rule Set compatibility as semantic work, not rule-file ingestion. This study follows a transaction through compilation and phased execution, then defines the conclusions supported by the available measurements.

01focused tests passed02clean request03rules / drops04CRS integration tests

Read the study→

Study / 03Published

Edge intelligence

11 July 2026

Learning without the hot path

A study of how Hyx AI reviews traffic, learns from outcomes, checks updates, and rolls back safely.

An evidence-controlled examination of Hyx AI as a complete production system rather than a standalone model. The study follows observations from request-time signals through a 62-feature contract, edge inference, asynchronous training, temporal validation, live shadowing, and reversible publication.

0162-feature contract02Record/score split03Temporal validation04Shadow promotion

Read the study→

Study / 02Published

Policy execution

11 July 2026

Policy at wire speed

A repository-backed study of compiled policy, semantic preservation, and microsecond-scale evaluation at production ruleset sizes.

An evidence-controlled examination of the Hyx Rules engine. The study traces policy from authored rules to an immutable execution artifact, measures representative and 3,000-rule workloads, and audits the correctness constraints around acceleration and live replacement.

01Release measurements02Semantic oracle03Optimizer safety04Live invalidation

Read the study→

Study / 01Published

Cache systems

11 July 2026

Mobius under pressure

A production-oriented study of scan resistance, bounded concurrency, and the measured gap to TinyUFO.

An evidence-controlled examination of Hyx’s in-memory cache engine. The study connects controlled internal measurements to repository invariants, then separates what the evidence establishes from what remains workload-dependent.

01Core-pinned comparison02Concurrent stress03Weighted bounds04Scan resistance

Read the study→